that are similar to those in humans. Using a hollow fibre system andS. aureus, Tam
et al. ( 2007 ) demonstrated a positive connection between the duration of garenox-
acin therapy and a decrease in the response to that antimicrobial agent. They
showed that doses producing an AUC 24 /MIC ratio of 114 h caused slightly greater
than a 5 log 10 (cfumLâ1) decline in the population if the administration was
extended for 96 h. However, if the same dose was administered for an additional
24 h or even longer, resistant populations began to emerge. They demonstrated that
the AUC 24 /MIC ratio necessary to suppress the resistant subpopulation over a
period of 48 h was 101 h whereas an AUC24/MIC ratio of 279 h was necessary
when using a full 10 days course of therapy. This reflects a nearly threefold increase
in drug exposure needed to minimise the selection of resistant bacteria.
The authors suggested that one reason for their observation may be the presence
of a pre-existent resistant sub-population. Amplification over time was a function of
drug pressure and of the change in the MIC caused by the mutation. The authors
also pointed out that, in the case of quinolones, another possible explanation is that
these agents interfere with DNA replication and cause surviving staphylococci to
undergo error-prone replication. Efflux-pump over-expression (by induction) may
partially account for the emergence of resistance. However, there may be an
interaction between these two mechanisms such that the early induction of efflux
pumps provided a small survivorship advantage and therefore time for additional
rounds of replication. Mutations can then arise from error-prone replication
mechanisms. As the duration of therapy increases, there is the risk of a greater
number of rounds of error-prone replication.
Although it may be considered appropriate to use findings such as these to
establish regulatory authority guidelines regarding the duration of therapy, Tam
et al. ( 2007 ) have pointed out that controlling drug dose and duration may not be
consistently helpful in the prevention of resistance. When DNA is transferred by
plasmids, transposons or transformation, it is unlikely that the selection of dose and
duration will have an important impact. This is also true when horizontal transmis-
sion of resistant microorganisms occurs. They do, however, point out that, in
circumstances in which dose and duration choice is helpful, resistance suppression
means that there will be no resistant microorganisms to transmit horizontally.
In response to these studies, Drusano et al. ( 2009 ) questioned the ultimate impact
of dose and duration on the relative proportions of wild-type and resistant bacteria
once therapy was discontinued. To address these concerns, they generated data on
the basis of four, five and six doses of the quinolone garenoxacin against the same
strain ofS. aureusused in the studies published by Tam et al. ( 2007 ). Observations
continued for more than one week after the final drug exposure.
Their study clearly showed that, in all instances, the rate of initial cell kill was
consistent with what had previously been observed, and the effect of drug on the
resistant population was similarly in agreement with the earlier work by Tam et al.
( 2007 ). When treatment was terminated, the total population continued to decline
for approximately 36 h, a finding attributed to slowly declining drug concentrations.
Thus, Drusano et al. ( 2009 ) demonstrated that even non-optimal dosing regimens
can be of therapeutic value if there is sufficient kill to allow host defences to
246 M. Martinez and P. Silley